1 .\" Automatically generated by Pod::Man 2.22 (Pod::Simple 3.10)
4 .\" ========================================================================
5 .de Sp \" Vertical space (when we can't use .PP)
9 .de Vb \" Begin verbatim text
14 .de Ve \" End verbatim text
18 .\" Set up some character translations and predefined strings. \*(-- will
19 .\" give an unbreakable dash, \*(PI will give pi, \*(L" will give a left
20 .\" double quote, and \*(R" will give a right double quote. \*(C+ will
21 .\" give a nicer C++. Capital omega is used to do unbreakable dashes and
22 .\" therefore won't be available. \*(C` and \*(C' expand to `' in nroff,
23 .\" nothing in troff, for use with C<>.
25 .ds C+ C\v'-.1v'\h'-1p'\s-2+\h'-1p'+\s0\v'.1v'\h'-1p'
29 . if (\n(.H=4u)&(1m=24u) .ds -- \(*W\h'-12u'\(*W\h'-12u'-\" diablo 10 pitch
30 . if (\n(.H=4u)&(1m=20u) .ds -- \(*W\h'-12u'\(*W\h'-8u'-\" diablo 12 pitch
43 .\" Escape single quotes in literal strings from groff's Unicode transform.
47 .\" If the F register is turned on, we'll generate index entries on stderr for
48 .\" titles (.TH), headers (.SH), subsections (.SS), items (.Ip), and index
49 .\" entries marked with X<> in POD. Of course, you'll have to process the
50 .\" output yourself in some meaningful fashion.
53 . tm Index:\\$1\t\\n%\t"\\$2"
63 .\" Accent mark definitions (@(#)ms.acc 1.5 88/02/08 SMI; from UCB 4.2).
64 .\" Fear. Run. Save yourself. No user-serviceable parts.
65 . \" fudge factors for nroff and troff
74 . ds #H ((1u-(\\\\n(.fu%2u))*.13m)
80 . \" simple accents for nroff and troff
90 . ds ' \\k:\h'-(\\n(.wu*8/10-\*(#H)'\'\h"|\\n:u"
91 . ds ` \\k:\h'-(\\n(.wu*8/10-\*(#H)'\`\h'|\\n:u'
92 . ds ^ \\k:\h'-(\\n(.wu*10/11-\*(#H)'^\h'|\\n:u'
93 . ds , \\k:\h'-(\\n(.wu*8/10)',\h'|\\n:u'
94 . ds ~ \\k:\h'-(\\n(.wu-\*(#H-.1m)'~\h'|\\n:u'
95 . ds / \\k:\h'-(\\n(.wu*8/10-\*(#H)'\z\(sl\h'|\\n:u'
97 . \" troff and (daisy-wheel) nroff accents
98 .ds : \\k:\h'-(\\n(.wu*8/10-\*(#H+.1m+\*(#F)'\v'-\*(#V'\z.\h'.2m+\*(#F'.\h'|\\n:u'\v'\*(#V'
99 .ds 8 \h'\*(#H'\(*b\h'-\*(#H'
100 .ds o \\k:\h'-(\\n(.wu+\w'\(de'u-\*(#H)/2u'\v'-.3n'\*(#[\z\(de\v'.3n'\h'|\\n:u'\*(#]
101 .ds d- \h'\*(#H'\(pd\h'-\w'~'u'\v'-.25m'\f2\(hy\fP\v'.25m'\h'-\*(#H'
102 .ds D- D\\k:\h'-\w'D'u'\v'-.11m'\z\(hy\v'.11m'\h'|\\n:u'
103 .ds th \*(#[\v'.3m'\s+1I\s-1\v'-.3m'\h'-(\w'I'u*2/3)'\s-1o\s+1\*(#]
104 .ds Th \*(#[\s+2I\s-2\h'-\w'I'u*3/5'\v'-.3m'o\v'.3m'\*(#]
105 .ds ae a\h'-(\w'a'u*4/10)'e
106 .ds Ae A\h'-(\w'A'u*4/10)'E
107 . \" corrections for vroff
108 .if v .ds ~ \\k:\h'-(\\n(.wu*9/10-\*(#H)'\s-2\u~\d\s+2\h'|\\n:u'
109 .if v .ds ^ \\k:\h'-(\\n(.wu*10/11-\*(#H)'\v'-.4m'^\v'.4m'\h'|\\n:u'
110 . \" for low resolution devices (crt and lpr)
111 .if \n(.H>23 .if \n(.V>19 \
124 .\" ========================================================================
126 .IX Title "PPI::Tokenizer 3"
127 .TH PPI::Tokenizer 3 "2009-08-08" "perl v5.8.7" "User Contributed Perl Documentation"
128 .\" For nroff, turn off justification. Always turn off hyphenation; it makes
129 .\" way too many mistakes in technical documents.
133 PPI::Tokenizer \- The Perl Document Tokenizer
135 .IX Header "SYNOPSIS"
137 \& # Create a tokenizer for a file, array or string
138 \& $Tokenizer = PPI::Tokenizer\->new( \*(Aqfilename.pl\*(Aq );
139 \& $Tokenizer = PPI::Tokenizer\->new( \e@lines );
140 \& $Tokenizer = PPI::Tokenizer\->new( \e$source );
142 \& # Return all the tokens for the document
143 \& my $tokens = $Tokenizer\->all_tokens;
145 \& # Or we can use it as an iterator
146 \& while ( my $Token = $Tokenizer\->get_token ) {
147 \& print "Found token \*(Aq$Token\*(Aq\en";
150 \& # If we REALLY need to manually nudge the cursor, you
151 \& # can do that to (The lexer needs this ability to do rollbacks)
152 \& $is_incremented = $Tokenizer\->increment_cursor;
153 \& $is_decremented = $Tokenizer\->decrement_cursor;
156 .IX Header "DESCRIPTION"
157 PPI::Tokenizer is the class that provides Tokenizer objects for use in
158 breaking strings of Perl source code into Tokens.
160 By the time you are reading this, you probably need to know a little
161 about the difference between how perl parses Perl \*(L"code\*(R" and how \s-1PPI\s0
162 parsers Perl \*(L"documents\*(R".
164 \&\*(L"perl\*(R" itself (the interpreter) uses a heavily modified lex specification
165 to specify its parsing logic, maintains several types of state as it
166 goes, and incrementally tokenizes, lexes \s-1AND\s0 \s-1EXECUTES\s0 at the same time.
168 In fact, it is provably impossible to use perl's parsing method without
169 simultaneously executing code. A formal mathematical proof has been
170 published demonstrating the method.
172 This is where the truism \*(L"Only perl can parse Perl\*(R" comes from.
174 \&\s-1PPI\s0 uses a completely different approach by abandoning the (impossible)
175 ability to parse Perl the same way that the interpreter does, and instead
176 parsing the source as a document, using a document structure independantly
177 derived from the Perl documentation and approximating the perl interpreter
178 interpretation as closely as possible.
180 It was touch and go for a long time whether we could get it close enough,
181 but in the end it turned out that it could be done.
183 In this approach, the tokenizer \f(CW\*(C`PPI::Tokenizer\*(C'\fR is implemented separately
184 from the lexer PPI::Lexer.
186 The job of \f(CW\*(C`PPI::Tokenizer\*(C'\fR is to take pure source as a string and break it
187 up into a stream/set of tokens, and contains most of the \*(L"black magic\*(R" used
188 in \s-1PPI\s0. By comparison, the lexer implements a relatively straight forward
189 tree structure, and has an implementation that is uncomplicated (compared
190 to the insanity in the tokenizer at least).
192 The Tokenizer uses an immense amount of heuristics, guessing and cruft,
193 supported by a very \fB\s-1VERY\s0\fR flexible internal \s-1API\s0, but fortunately it was
194 possible to largely encapsulate the black magic, so there is not a lot that
195 gets exposed to people using the \f(CW\*(C`PPI::Tokenizer\*(C'\fR itself.
198 Despite the incredible complexity, the Tokenizer itself only exposes a
199 relatively small number of methods, with most of the complexity implemented
201 .ie n .SS "new $file | \e@lines | \e$source"
202 .el .SS "new \f(CW$file\fP | \e@lines | \e$source"
203 .IX Subsection "new $file | @lines | $source"
204 The main \f(CW\*(C`new\*(C'\fR constructor creates a new Tokenizer object. These
205 objects have no configuration parameters, and can only be used once,
206 to tokenize a single perl source file.
208 It takes as argument either a normal scalar containing source code,
209 a reference to a scalar containing source code, or a reference to an
210 \&\s-1ARRAY\s0 containing newline-terminated lines of source code.
212 Returns a new \f(CW\*(C`PPI::Tokenizer\*(C'\fR object on success, or throws a
213 PPI::Exception exception on error.
215 .IX Subsection "get_token"
216 When using the PPI::Tokenizer object as an iterator, the \f(CW\*(C`get_token\*(C'\fR
217 method is the primary method that is used. It increments the cursor
218 and returns the next Token in the output array.
220 The actual parsing of the file is done only as-needed, and a line at
221 a time. When \f(CW\*(C`get_token\*(C'\fR hits the end of the token array, it will
222 cause the parser to pull in the next line and parse it, continuing
223 as needed until there are more tokens on the output array that
224 get_token can then return.
226 This means that a number of Tokenizer objects can be created, and
227 won't consume significant \s-1CPU\s0 until you actually begin to pull tokens
230 Return a PPI::Token object on success, \f(CW0\fR if the Tokenizer had
231 reached the end of the file, or \f(CW\*(C`undef\*(C'\fR on error.
233 .IX Subsection "all_tokens"
234 When not being used as an iterator, the \f(CW\*(C`all_tokens\*(C'\fR method tells
235 the Tokenizer to parse the entire file and return all of the tokens
236 in a single \s-1ARRAY\s0 reference.
238 It should be noted that \f(CW\*(C`all_tokens\*(C'\fR does \fB\s-1NOT\s0\fR interfere with the
239 use of the Tokenizer object as an iterator (does not modify the token
240 cursor) and use of the two different mechanisms can be mixed safely.
242 Returns a reference to an \s-1ARRAY\s0 of PPI::Token objects on success
243 or throws an exception on error.
244 .SS "increment_cursor"
245 .IX Subsection "increment_cursor"
246 Although exposed as a public method, \f(CW\*(C`increment_method\*(C'\fR is implemented
247 for expert use only, when writing lexers or other components that work
248 directly on token streams.
250 It manually increments the token cursor forward through the file, in effect
251 \&\*(L"skipping\*(R" the next token.
253 Return true if the cursor is incremented, \f(CW0\fR if already at the end of
254 the file, or \f(CW\*(C`undef\*(C'\fR on error.
255 .SS "decrement_cursor"
256 .IX Subsection "decrement_cursor"
257 Although exposed as a public method, \f(CW\*(C`decrement_method\*(C'\fR is implemented
258 for expert use only, when writing lexers or other components that work
259 directly on token streams.
261 It manually decrements the token cursor backwards through the file, in
262 effect \*(L"rolling back\*(R" the token stream. And indeed that is what it is
263 primarily intended for, when the component that is consuming the token
264 stream needs to implement some sort of \*(L"roll back\*(R" feature in its use
267 Return true if the cursor is decremented, \f(CW0\fR if already at the
268 beginning of the file, or \f(CW\*(C`undef\*(C'\fR on error.
271 .SS "How the Tokenizer Works"
272 .IX Subsection "How the Tokenizer Works"
273 Understanding the Tokenizer is not for the feint-hearted. It is by far
274 the most complex and twisty piece of perl I've ever written that is actually
275 still built properly and isn't a terrible spaghetti-like mess. In fact, you
276 probably want to skip this section.
278 But if you really want to understand, well then here goes.
279 .SS "Source Input and Clean Up"
280 .IX Subsection "Source Input and Clean Up"
281 The Tokenizer starts by taking source in a variety of forms, sucking it
282 all in and merging into one big string, and doing our own internal line
283 split, using a \*(L"universal line separator\*(R" which allows the Tokenizer to
284 take source for any platform (and even supports a few known types of
285 broken newlines caused by mixed mac/pc/*nix editor screw ups).
287 The resulting array of lines is used to feed the tokenizer, and is also
288 accessed directly by the heredoc-logic to do the line-oriented part of
290 .SS "Doing Things the Old Fashioned Way"
291 .IX Subsection "Doing Things the Old Fashioned Way"
292 Due to the complexity of perl, and after 2 previously aborted parser
293 attempts, in the end the tokenizer was fashioned around a line-buffered
294 character-by-character method.
296 That is, the Tokenizer pulls and holds a line at a time into a line buffer,
297 and then iterates a cursor along it. At each cursor position, a method is
298 called in whatever token class we are currently in, which will examine the
299 character at the current position, and handle it.
301 As the handler methods in the various token classes are called, they
302 build up a output token array for the source code.
304 Various parts of the Tokenizer use look-ahead, arbitrary-distance
305 look-behind (although currently the maximum is three significant tokens),
306 or both, and various other heuristic guesses.
308 I've been told it is officially termed a \fI\*(L"backtracking parser
309 with infinite lookaheads\*(R"\fR.
310 .SS "State Variables"
311 .IX Subsection "State Variables"
312 Aside from the current line and the character cursor, the Tokenizer
313 maintains a number of different state variables.
314 .IP "Current Class" 4
315 .IX Item "Current Class"
316 The Tokenizer maintains the current token class at all times. Much of the
317 time is just going to be the \*(L"Whitespace\*(R" class, which is what the base of
318 a document is. As the tokenizer executes the various character handlers,
319 the class changes a lot as it moves a long. In fact, in some instances,
320 the character handler may not handle the character directly itself, but
321 rather change the \*(L"current class\*(R" and then hand off to the character
322 handler for the new class.
324 Because of this, and some other things I'll deal with later, the number of
325 times the character handlers are called does not in fact have a direct
326 relationship to the number of actual characters in the document.
328 .IX Item "Current Zone"
329 Rather than create a class stack to allow for infinitely nested layers of
330 classes, the Tokenizer recognises just a single layer.
332 To put it a different way, in various parts of the file, the Tokenizer will
333 recognise different \*(L"base\*(R" or \*(L"substrate\*(R" classes. When a Token such as a
334 comment or a number is finalised by the tokenizer, it \*(L"falls back\*(R" to the
337 This allows proper tokenization of special areas such as _\|_DATA_\|_
338 and _\|_END_\|_ blocks, which also contain things like comments and \s-1POD\s0,
339 without allowing the creation of any significant Tokens inside these areas.
341 For the main part of a document we use PPI::Token::Whitespace for this,
342 with the idea being that code is \*(L"floating in a sea of whitespace\*(R".
343 .IP "Current Token" 4
344 .IX Item "Current Token"
345 The final main state variable is the \*(L"current token\*(R". This is the Token
346 that is currently being built by the Tokenizer. For certain types, it
347 can be manipulated and morphed and change class quite a bit while being
348 assembled, as the Tokenizer's understanding of the token content changes.
350 When the Tokenizer is confident that it has seen the end of the Token, it
351 will be \*(L"finalized\*(R", which adds it to the output token array and resets
352 the current class to that of the zone that we are currently in.
354 I should also note at this point that the \*(L"current token\*(R" variable is
355 optional. The Tokenizer is capable of knowing what class it is currently
356 set to, without actually having accumulated any characters in the Token.
357 .SS "Making It Faster"
358 .IX Subsection "Making It Faster"
359 As I'm sure you can imagine, calling several different methods for each
360 character and running regexes and other complex heuristics made the first
361 fully working version of the tokenizer extremely slow.
363 During testing, I created a metric to measure parsing speed called
364 \&\s-1LPGC\s0, or \*(L"lines per gigacycle\*(R" . A gigacycle is simple a billion \s-1CPU\s0
365 cycles on a typical single-core \s-1CPU\s0, and so a Tokenizer running at
366 \&\*(L"1000 lines per gigacycle\*(R" should generate around 1200 lines of tokenized
367 code when running on a 1200 MHz processor.
369 The first working version of the tokenizer ran at only 350 \s-1LPGC\s0, so to
370 tokenize a typical large module such as ExtUtils::MakeMaker took
371 10\-15 seconds. This sluggishness made it unpractical for many uses.
373 So in the current parser, there are multiple layers of optimisation
374 very carefully built in to the basic. This has brought the tokenizer
375 up to a more reasonable 1000 \s-1LPGC\s0, at the expense of making the code
376 quite a bit twistier.
377 .SS "Making It Faster \- Whole Line Classification"
378 .IX Subsection "Making It Faster - Whole Line Classification"
379 The first step in the optimisation process was to add a hew handler to
380 enable several of the more basic classes (whitespace, comments) to be
381 able to be parsed a line at a time. At the start of each line, a
382 special optional handler (only supported by a few classes) is called to
383 check and see if the entire line can be parsed in one go.
385 This is used mainly to handle things like \s-1POD\s0, comments, empty lines,
386 and a few other minor special cases.
387 .SS "Making It Faster \- Inlining"
388 .IX Subsection "Making It Faster - Inlining"
389 The second stage of the optimisation involved inlining a small
390 number of critical methods that were repeated an extremely high number
391 of times. Profiling suggested that there were about 1,000,000 individual
392 method calls per gigacycle, and by cutting these by two thirds a significant
393 speed improvement was gained, in the order of about 50%.
395 You may notice that many methods in the \f(CW\*(C`PPI::Tokenizer\*(C'\fR code look
396 very nested and long hand. This is primarily due to this inlining.
398 At around this time, some statistics code that existed in the early
399 versions of the parser was also removed, as it was determined that
400 it was consuming around 15% of the \s-1CPU\s0 for the entire parser, while
401 making the core more complicated.
403 A judgment call was made that with the difficulties likely to be
404 encountered with future planned enhancements, and given the relatively
405 high cost involved, the statistics features would be removed from the
407 .SS "Making It Faster \- Quote Engine"
408 .IX Subsection "Making It Faster - Quote Engine"
409 Once inlining had reached diminishing returns, it became obvious from
410 the profiling results that a huge amount of time was being spent
411 stepping a char at a time though long, simple and \*(L"syntactically boring\*(R"
412 code such as comments and strings.
414 The existing regex engine was expanded to also encompass quotes and
415 other quote-like things, and a special abstract base class was added
416 that provided a number of specialised parsing methods that would \*(L"scan
417 ahead\*(R", looking out ahead to find the end of a string, and updating
418 the cursor to leave it in a valid position for the next call.
420 This is also the point at which the number of character handler calls began
421 to greatly differ from the number of characters. But it has been done
422 in a way that allows the parser to retain the power of the original
423 version at the critical points, while skipping through the \*(L"boring bits\*(R"
424 as needed for additional speed.
426 The addition of this feature allowed the tokenizer to exceed 1000 \s-1LPGC\s0
428 .ie n .SS "Making It Faster \- The ""Complete"" Mechanism"
429 .el .SS "Making It Faster \- The ``Complete'' Mechanism"
430 .IX Subsection "Making It Faster - The Complete Mechanism"
431 As it became evident that great speed increases were available by using
432 this \*(L"skipping ahead\*(R" mechanism, a new handler method was added that
433 explicitly handles the parsing of an entire token, where the structure
434 of the token is relatively simple. Tokens such as symbols fit this case,
435 as once we are passed the initial sigil and word char, we know that we
436 can skip ahead and \*(L"complete\*(R" the rest of the token much more easily.
438 A number of these have been added for most or possibly all of the common
439 cases, with most of these \*(L"complete\*(R" handlers implemented using regular
442 In fact, so many have been added that at this point, you could arguably
443 reclassify the tokenizer as a \*(L"hybrid regex, char\-by=char heuristic
444 tokenizer\*(R". More tokens are now consumed in \*(L"complete\*(R" methods in a
445 typical program than are handled by the normal char-by-char methods.
447 Many of the these complete-handlers were implemented during the writing
448 of the Lexer, and this has allowed the full parser to maintain around
449 1000 \s-1LPGC\s0 despite the increasing weight of the Lexer.
450 .SS "Making It Faster \- Porting To C (In Progress)"
451 .IX Subsection "Making It Faster - Porting To C (In Progress)"
452 While it would be extraordinarily difficult to port all of the Tokenizer
453 to C, work has started on a \s-1PPI::XS\s0 \*(L"accelerator\*(R" package which acts as
454 a separate and automatically-detected add-on to the main \s-1PPI\s0 package.
456 \&\s-1PPI::XS\s0 implements faster versions of a variety of functions scattered
457 over the entire \s-1PPI\s0 codebase, from the Tokenizer Core, Quote Engine, and
458 various other places, and implements them identically in \s-1XS/C\s0.
460 In particular, the skip-ahead methods from the Quote Engine would appear
461 to be extremely amenable to being done in C, and a number of other
462 functions could be cherry-picked one at a time and implemented in C.
464 Each method is heavily tested to ensure that the functionality is
465 identical, and a versioning mechanism is included to ensure that if a
466 function gets out of sync, \s-1PPI::XS\s0 will degrade gracefully and just
467 not replace that single method.
470 \&\- Add an option to reset or seek the token stream...
472 \&\- Implement more Tokenizer functions in \s-1PPI::XS\s0
475 See the support section in the main module.
478 Adam Kennedy <adamk@cpan.org>
480 .IX Header "COPYRIGHT"
481 Copyright 2001 \- 2009 Adam Kennedy.
483 This program is free software; you can redistribute
484 it and/or modify it under the same terms as Perl itself.
486 The full text of the license can be found in the
487 \&\s-1LICENSE\s0 file included with this module.
projects / catagits/Gitalist.git / blob